Entries
Architecture, as announced, goes;
β’ Multi agent AI runtime
β’ Custom desktop shell
β’ Linux base for x86_64 and ARM64
β’ systemd, PipeWire, Mesa
β’ Node.js 22, Python 3, SQLite
β’ Web MCP integration
β’ KVM acceleration on Linux
β’ Apple Silicon support via QEMU HVF
β’ Fully buildable and reproducible from source
The key difference is architectural.
The AI is not an application running inside the OS.
It is part of the operating system itself.
The goal is simple: reduce friction between intent and execution at the system level.
A native operating system for agents work. That claims to have given direct access to ai agents on the system. An interesting approach, but no Github repro, no link, no nothing. Instead, they require users to dm on reddit. May be a closed beta, may not; don't yet know.
Anyway, the idea deserves a shot.